docs: replace README with Next.js boilerplate template

Updated README.md from a detailed OllaHub project description to the standard Next.js getting started template, likely to initialize or reset the project as a Next.js application.

lib/chat.ts

@@ -0,0 +1,50 @@
+export type ChatRole = "user" | "assistant" | "system";
+
+export type Message = {
+  id: string;
+  role: ChatRole;
+  content: string;
+  timestamp: number;
+};
+
+export type ChatState = {
+  messages: Message[];
+  selectedModel: string | null;
+  isStreaming: boolean;
+  streamingContent: string;
+};
+
+export function createMessage(role: ChatRole, content: string): Message {
+  return {
+    id: crypto.randomUUID(),
+    role,
+    content,
+    timestamp: Date.now(),
+  };
+}
+
+export function formatTimestamp(timestamp: number): string {
+  const now = Date.now();
+  const diff = now - timestamp;
+  const minutes = Math.floor(diff / 60000);
+
+  if (minutes < 1) return "agora";
+  if (minutes < 60) return `${minutes} min atrás`;
+
+  const hours = Math.floor(minutes / 60);
+  if (hours < 24) return `${hours}h atrás`;
+
+  const days = Math.floor(hours / 24);
+  return `${days}d atrás`;
+}
+
+export function convertToOllamaMessages(messages: Message[]) {
+  return messages.map((msg) => ({
+    role: msg.role,
+    content: msg.content,
+  }));
+}
+
+export function createSystemMessage(content: string): Message {
+  return createMessage("system", content);
+}

lib/models.ts

@@ -0,0 +1,100 @@
+export type Device = "CPU" | "GPU";
+
+export type ModelInfo = {
+  id: string;
+  name: string;
+  sizeGB: number;
+  quantization: string;
+  device: Device;
+  estCpuUsagePct?: number;
+  estVramGB?: number;
+};
+
+export const mockModels: ModelInfo[] = [
+  {
+    id: "llama3.1-8b-q4",
+    name: "llama3.1:8b-q4",
+    sizeGB: 4.8,
+    quantization: "Q4",
+    device: "CPU",
+    estCpuUsagePct: 85,
+    estVramGB: 0.0,
+  },
+  {
+    id: "llama3.1-8b-q8",
+    name: "llama3.1:8b-q8",
+    sizeGB: 8.9,
+    quantization: "Q8",
+    device: "GPU",
+    estCpuUsagePct: 20,
+    estVramGB: 8,
+  },
+  {
+    id: "mistral-7b-q4",
+    name: "mistral:7b-q4",
+    sizeGB: 4.1,
+    quantization: "Q4",
+    device: "CPU",
+    estCpuUsagePct: 75,
+    estVramGB: 0,
+  },
+  {
+    id: "mistral-7b-q8",
+    name: "mistral:7b-q8",
+    sizeGB: 7.9,
+    quantization: "Q8",
+    device: "GPU",
+    estCpuUsagePct: 18,
+    estVramGB: 8,
+  },
+  {
+    id: "qwen2.5-7b-q4",
+    name: "qwen2.5:7b-q4",
+    sizeGB: 4.2,
+    quantization: "Q4",
+    device: "CPU",
+    estCpuUsagePct: 80,
+    estVramGB: 0,
+  },
+  {
+    id: "qwen2.5-7b-q8",
+    name: "qwen2.5:7b-q8",
+    sizeGB: 8.1,
+    quantization: "Q8",
+    device: "GPU",
+    estCpuUsagePct: 22,
+    estVramGB: 8,
+  },
+];
+
+export function inferQuantization(nameOrDetails: string | unknown): string {
+  const s =
+    typeof nameOrDetails === "string" ? nameOrDetails.toLowerCase() : "";
+  if (s.includes("q8")) return "Q8";
+  if (s.includes("q5")) return "Q5";
+  if (s.includes("q4")) return "Q4";
+  return "Q?";
+}
+
+export function inferDevice(details?: unknown): Device {
+  const s = typeof details === "string" ? details.toLowerCase() : "";
+  if (s.includes("cuda") || s.includes("metal") || s.includes("gpu"))
+    return "GPU";
+  return "CPU";
+}
+
+export function estimateCpuVram(
+  info: Pick<ModelInfo, "sizeGB" | "quantization" | "device">,
+) {
+  const q = info.quantization.toUpperCase();
+  let cpuPct = 50;
+  let vramGB = 0;
+  if (info.device === "GPU") {
+    vramGB = Math.max(4, Math.round(info.sizeGB));
+    cpuPct = 20;
+  } else {
+    // heurística simples: quantizações mais baixas usam mais CPU
+    cpuPct = q === "Q8" ? 60 : q === "Q5" ? 70 : 85;
+  }
+  return { cpuPct, vramGB };
+}

lib/ollama.ts

@@ -0,0 +1,42 @@
+import { Ollama } from "ollama";
+
+export const OLLAMA_HOST = process.env.OLLAMA_HOST || "http://localhost:11434";
+
+export async function listModelsViaSdk() {
+  try {
+    const client = new Ollama({ host: OLLAMA_HOST });
+    const res = await client.list();
+    return res?.models ?? [];
+  } catch (e) {
+    throw e;
+  }
+}
+
+export async function listModelsViaHttp() {
+  const url = `${OLLAMA_HOST}/api/tags`;
+  const res = await fetch(url, { cache: "no-store" });
+  if (!res.ok) throw new Error(`Ollama HTTP error ${res.status}`);
+  const data = await res.json();
+  return data?.models ?? [];
+}
+
+export async function chatWithStream(
+  model: string,
+  messages: Array<{ role: string; content: string }>,
+  options?: Record<string, unknown>,
+  system?: string
+) {
+  const client = new Ollama({ host: OLLAMA_HOST });
+  return client.chat({
+    model,
+    messages,
+    stream: true,
+    options,
+    system,
+  });
+}
+
+export async function pullModel(model: string) {
+  const client = new Ollama({ host: OLLAMA_HOST });
+  return client.pull({ model, stream: true });
+}

lib/systemPrompt.ts

@@ -0,0 +1,30 @@
+export const DEFAULT_SYSTEM_PROMPT = `Você é um assistente de IA especializado em desenvolvimento de software e análise técnica.
+
+FORMATAÇÃO OBRIGATÓRIA:
+- Sempre use blocos de código markdown com especificação de linguagem
+- Para JavaScript/TypeScript: \`\`\`javascript ou \`\`\`typescript
+- Para Python: \`\`\`python
+- Para Bash/Shell: \`\`\`bash
+- Para SQL: \`\`\`sql
+- Para JSON: \`\`\`json
+- Para YAML: \`\`\`yaml
+- Para Docker: \`\`\`dockerfile
+- Para configurações: \`\`\`toml, \`\`\`ini, \`\`\`conf
+
+ESTRUTURA DE RESPOSTAS:
+1. Explicação conceitual clara
+2. Exemplo prático em bloco de código
+3. Explicação do código quando necessário
+4. Boas práticas
+5. Possíveis problemas e soluções
+
+CONTEXTO OLLAHUB:
+- Execução local via Ollama
+- Stack frequente: Next.js, React, Node.js
+- Priorize exemplos executáveis e explique comandos e configurações.`;
+
+export function mergeSystemPrompt(userPrompt?: string) {
+  if (!userPrompt || userPrompt.trim().length === 0)
+    return DEFAULT_SYSTEM_PROMPT;
+  return `${DEFAULT_SYSTEM_PROMPT}\n\nDiretivas do usuário:\n${userPrompt}`;
+}